<damo22> mrnuke: !!
<damo22> mrkiko: yes, i will get my multimeter back from lending it out, tomorrow
<damo22> mrnuke: i found a lenovo g505s with A10-5750M APU for USD$35 on ebay the other day
<damo22> shipping was way more than the unit itself, but i really want one again
<damo22> ive never ported a router before, if the chipset is supported, it must be pretty easy given the hardware is almost fully described in a dts file?
Emantor has quit [Quit: ZNC - http://znc.in]
Emantor has joined #openwrt-devel
<djfe> I just noticed that iperf3 reproducibly segfaults on certain devices https://github.com/openwrt/packages/issues/24116
<djfe> (probably the whole ath79-generic target)
<djfe> 22.03 and 23.05 appear to be affected
minimal has quit [Quit: Leaving]
<mrnuke> damo22: Did we ever figure out a SMU replacement for those? I think they were before PSP came out in AMD silicon
tSYS has quit [Quit: *squeak*]
tSYS has joined #openwrt-devel
<mrnuke> damo22: BTW, router porting is much easier and fun. I have a few APs and switches under my belt. Someone left their IP TV on loud? `ifconfig eth<n> down`. It's lovely
<damo22> mrnuke: yes, https://github.com/zamaudio/smutool firmware runs on SMU on f2a85m and keeps the temp at 45C in memtest.... i think it works, but im not convinced i would replace it
<damo22> SMU is basically a PCI mailbox that pokes at registers that are undocumented on AMD, but you could probably poke them on PCI manually if you knew what to poke
<damo22> but it does have an out of band microcontroller that runs a loop monitoring the temperature and setting the fan speed
<damo22> i think they tried to hide the interesting power registers from the docs and put a microcontroller in front of them with some kind of API, but the underlying regs are still there, just, undocumented...
goliath has quit [Quit: SIGSEGV]
<damo22> if i manage to port this 4g modem/router, i will be able to decommission my old phone which is currently tethered to my old router
ssterling has joined #openwrt-devel
<mrnuke> damo22: Nice!
Mangix has quit [Ping timeout: 480 seconds]
<ssterling> I’m trying to port to a new device, but the build is failing on target/linux with this message: “cat: /home/smp/programs/openwrt/build_dir/target-arm_cortex-a9_musl_eabi/linux-bcm53xx_generic/image-.dtb: No such file or directory”
<ssterling> I have a vague understanding of device trees, and I think I know where this message is coming from, but I’m kinda stumped
rz has quit [Remote host closed the connection]
rz has joined #openwrt-devel
Mangix has joined #openwrt-devel
<ynezz> Ansuel: mrnuke: stintel: just for the record, opened it and Zyxel XMG1915-10EP is RTL9302
<ynezz> svanheule: just wanted to add: | Zyxel | [[XMG1915-10EP]] | 8x 2.5GbE, 2 x 10G SFP+, PoE++ (60W) | [[RTL93xx|RTL9302]] ||
<ynezz> for some dokuwiki reasons it doesn't work and I give up :P
<owrt-images-builds> Build [#73](https://buildbot.openwrt.org/images/#/builders/108/builds/73) of `openwrt-23.05_oxnas/ox820` failed.
<svanheule> ynezz: I'll have a look
<ynezz> it works fine on openwrt.org, so probably something in your plugins
<ynezz> triggered by either that '++' or '(60W)'
rua is now known as Guest5798
rua has joined #openwrt-devel
<ynezz> it seems, that it even don't work with ''8x 2.5GbE, 2 x 10G SFP+, PoE++ (60W)''
<ynezz> every dokuwiki interaction feels like teleporting 20 years back :P
rua has quit [Quit: Leaving.]
<svanheule> the '++' is causing issues because of the folded plugin
<stintel> speaking about Realtek, I made a temporary fiber link between the two floors in my house using BiDir SFPs
<stintel> on the ECS4100-12PH
<stintel> it was a bit of a gamble, had no idea if it was going to work
<svanheule> ynezz: I've removed the folded plugin and updated the switch model table to contain '++' where appropriate
<svanheule> stintel: you have it working then?
<stintel> aye!
<stintel> been meaning to reply to Andreas' mail but didn't find the time
gladiac is now known as Guest5815
gladiac has joined #openwrt-devel
Guest5815 has quit [Ping timeout: 480 seconds]
<KGB-2> https://tests.reproducible-builds.org/openwrt/openwrt_mpc85xx.html has been updated. (100.0% images and 98.3% packages reproducible in our current test framework.)
stklcode has joined #openwrt-devel
robimarko has joined #openwrt-devel
clandmeter has quit [Quit: Alpine Linux, the security-oriented, lightweight Linux distribution]
rua has joined #openwrt-devel
clandmeter has joined #openwrt-devel
wille-io has quit [Ping timeout: 480 seconds]
rua has quit [Quit: Leaving.]
rua has joined #openwrt-devel
wille-io has joined #openwrt-devel
rua has quit [Quit: Leaving.]
stklcode has quit [Ping timeout: 480 seconds]
* russell-- just tried flashing a purported ubnt nanostation m xw and the ethernet didn't work
<russell--> build from 2019-03-01 works
robimarko has quit [Ping timeout: 480 seconds]
<russell--> ah, i figured it out: switch configuration problem
<russell--> all better now
landswipe has joined #openwrt-devel
<landswipe> hi everyone, can someone point me to an example where a custom kmod package can be compiled from a package that I have in my own feed (via the sdk)?
<landswipe> I want to get a build of https://github.com/frank-zago/ch341-i2c-spi-gpio.git for my device.
robimarko has joined #openwrt-devel
rsalvaterra has joined #openwrt-devel
<russell--> lol, now a ubnt_nanostation-m's eth0 doesn't seem to work
minimal has joined #openwrt-devel
tidalf has joined #openwrt-devel
landswipe has quit [Ping timeout: 480 seconds]
tidalf has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tidalf has joined #openwrt-devel
tidalf has quit []
tidalf has joined #openwrt-devel
tidalf has quit []
tidalf has joined #openwrt-devel
<russell--> looks like eth0 and eth1 got swapped
<russell--> since 2019 when i last flashed this thing
tidalf has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tidalf has joined #openwrt-devel
f00b4r0 has quit [Quit: Textual IRC Client: www.textualapp.com]
rua has joined #openwrt-devel
rua has quit [Remote host closed the connection]
goliath has joined #openwrt-devel
<Ansuel> Mangix https://gist.github.com/Ansuel/1575378954a0aee5a4e53ffa3ab5084f hope it does apply and compile well but should give us some clue if something is not applying right
Snuupy has quit [Quit: leaving]
valku has joined #openwrt-devel
tidalf has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tidalf has joined #openwrt-devel
Mangix has quit [Read error: Connection reset by peer]
c512l has joined #openwrt-devel
tidalf is now known as Guest5923
Guest5923 has quit [Ping timeout: 480 seconds]
stklcode has joined #openwrt-devel
stklcode has quit [Quit: Leaving]
Snuupy has joined #openwrt-devel
Acinonyx has joined #openwrt-devel
Acinonyx has quit [Read error: Connection reset by peer]
Mangix has joined #openwrt-devel
<Mangix> Ansuel: alright
<Mangix> Ansuel: both hunks failed to apply. kernel 6.6
<Mangix> hrm even with 6.1
tidalf has joined #openwrt-devel
minimal has quit [Quit: Leaving]
<Mangix> Ansuel: still no
<Ansuel> compilation error ?
<Mangix> drivers/net/dsa/qca/qca8k-8xxx.c:514:60: error: expected ';' before 'val_buf2'
<Mangix> lol
<Ansuel> :D
<Ansuel> retry
<Mangix> error: implicit declaration of function 'malloc' [-Wimplicit-function-declaration]
<Mangix> shouldn't this be kmalloc?
<Mangix> 525 | if (val_buf[i] != val_buf2[i])
<Mangix> error: void value not ignored as it ought to be
robimarko has quit [Ping timeout: 480 seconds]
<Ansuel> love when i make patch on hurry :D
<Ansuel> updated again
<Mangix> no way that works
<Mangix> val_buf is a void pointer
<Mangix> i fixed it up locally
<Ansuel> cast to u32
<Mangix> I added u32 *val = (u32 *)val_buf; as in the following function
<Ansuel> yep it's ok
<Mangix> so this logs to dmesg?
<Ansuel> yes
<Ansuel> if something is different
<Ansuel> ok run a bridge fdb show
<Ansuel> and bridge vlan show
<Ansuel> anything logged?
<Ansuel> mhh the write diff are O.K. since we write in the ATU and the reg changes
<Ansuel> sooo mgmt eth is working correctly...
<Ansuel> the patch disable mgmt eth for phy we need to check if ports die after 30 seconds with this patch
<Mangix> doesn't that effectively fix the issue? That's what the if 0 patch does
<Mangix> and no, LAN ports are still operational
<Ansuel> the if 0 patch disables mgmt eth globally
<Ansuel> the current patch disable it only setting and reading reg... but phy operation use the old way
<Mangix> That's in qca8k_internal_mdio_read
<Mangix> which means your patch effectively fixes whatever issue there is
<Ansuel> we should check if reg read differs there...
<Ansuel> also I notice something strange
<Mangix> hmm?
<Ansuel> inspecting your packet
<Ansuel> the seq was always a big number...
<Ansuel> i wonder if there is a problem there
<Ansuel> in qca8k_mdio_header_fill_seq_num
<Ansuel> add pr_info("SEQ %x\n", seq_num);
<Ansuel> and also uncomment stuff in internal mdio read and write
<Ansuel> you should get dmesg spammed with this stuff
<Mangix> fun
<Mangix> [ 77.072200] REG 430 write diff MGMT_ETH: 10001 MII: f97cce9f
<Mangix> looks funky
<Ansuel> wait this wasn't in the original dmesg
<Mangix> yeah it was
<Ansuel> ok notice
<Ansuel> yes it's strange
<Ansuel> let me check what is reg 430
<Mangix> what the...
<Mangix> [ 1552.391076] REG 60c write diff MGMT_ETH: 80000003 MII: 3
<Mangix> [ 1552.397108] REG 60c write diff MGMT_ETH: 80000003 MII: 1003
<Ansuel> those are ATU and are ok
<Ansuel> the bit 31 is AT_BUSY
<Ansuel> that is hw reset
<Mangix> Ansuel: SEQ goes from 1 to f
<Ansuel> reg 430 is vlan default setting
<Mangix> otherwise it's sequential
<Ansuel> 1 to f ?
<Mangix> this spam is crazy
<Mangix> it remains sequential
<Ansuel> ok and every port works currently right?
<Mangix> for now
<Ansuel> do you have a way to check when they stop working?
<Mangix> currently 280 in dmesg
<Ansuel> might be interesting to check the sequence number
<Mangix> last time it failed was at 4100
<Ansuel> the seq is 32bit
<Mangix> just went past 8000
<Mangix> wonder if the issue happens when it goes past ffff ...
<Ansuel> yes i want to check that
<Ansuel> bte reg 90 is ok to have different values
<Mangix> alright, went past ffff just fine
<Mangix> lan4 is still u[p
<Mangix> *up
<Ansuel> after ffff it did roll to 0 right?
<Mangix> no
<Mangix> 10000
<Mangix> it's at 12000 now
<Ansuel> max is FFFFFFFF
<Mangix> alright, back to mwlwifi while waitingb
<Ansuel> what is wrong with mwlwifi ?
<Mangix> 64-bit compilation errors
<Mangix> apparently my understanding of C is awful.
<Ansuel> mh?
<Ansuel> and the error you are fixing is?
<Mangix> error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
<Ansuel> Mangix *((uintptr_t *)le32_to_cpu(get_hw_spec->wcb_base0)) ?
<Mangix> IIRC I tried that.
<Mangix> uh oh
<Ansuel> ?
<Mangix> so lan4 went down
<Mangix> unfortunately it seems I missed when it did
<Ansuel> seq is ?
<Mangix> when I stopped it, it was [ 2908.824416] SEQ 394bd
<Mangix> but it failed earlier than that
<Ansuel> and no ports works now?
<Mangix> I''m guessing it's not seq related then
<Ansuel> does bridge fdb show works?
<Mangix> all ports except 2
<Mangix> I rebooted unfortunately
<Mangix> I'll try timing it this time
<Ansuel> i need to understand if the entire mgmt eth blow up or something else
<Mangix> good question.
<Mangix> the PHY is working
<Mangix> plugging and unplugging has LEDs turning on and off
<Ansuel> one thing i can check is compare a dump
<Ansuel> between my ipq806x and yours just to make sure there isn't something wrong in how the packet is crafted
<Ansuel> cause it really looks like after a while packets are ignored
<Mangix> Ansuel: uintptr_t does not fix it
<Mangix> Ansuel: before and after lan4 fails: https://gist.github.com/neheb/80f5ca056009e4e8a4d7591765ac085f
<Mangix> Ansuel: hrm I don't think the counter is related
<Mangix> this time it failed at around f6c0
<Mangix> before it went to ~30000
<Mangix> After it failed I also see [ 783.361315] REG 60c write diff MGMT_ETH: 80004505 MII: 4505
<Mangix> hrm so I made a small C example of the mwlwifi in question. segfaults. of course.